add support for zipping a sequence of RDDs

[mohitjaggi]

a proposed fix for https://issues.apache.org/jira/browse/SPARK-3489

[SparkQA]

Can one of the admins verify this patch?

[srowen]

ok to test

[srowen]

(This would need a JIRA, if it proceeds.) How strong is the need for this? you can always zip several RDDs together directly with several calls. It gets unwieldy with a lot of RDDs, but do you need to do that? you lose type safety in this method. I wonder if, is there a common need to zip an RDD with 2-3 others, we can just add utility overloads for that, that preserve type safety too.

[SparkQA]

Test build #27098 has finished for PR 2429 at commit 21bdd19.

• This patch fails Spark unit tests.
• This patch merges cleanly.
• This patch adds no public classes.

[JoshRosen]

@srowen We already have versions of zipPartitions for zipping together up to 4 RDDs.

Is there a common enough use case for this feature that we would want to expose this non-type-safe API?

[srowen]

Ah right, I was looking at zip rather than zipPartitions. Yes the overloads are already there for zipPartitions and I don't see value in adding more. At best, add overloads for zip? but I also wonder about the use case, since you can still achieve this if really needed by calling zipPartitions overloads to make your own zip.

[AmplabJenkins]

Can one of the admins verify this patch?

[srowen]

Do you mind closing this PR?

[mohitjaggi]

closing on sean's request. i have a workaround.